Re: st: two-way fixed effects

[David Jacobs <jacobs.184@sociology.osu.edu>]

First, you probably should estimate with -xtreg- particularly if you 
are using robust or cluster to correct your standard errors.  This is 
the designated fixed-effects procedure in Stata and it uses more 
appropriate cluster or robust corrections for the standard 
errors.  Note that you will still need to include the time-specific 
dummies as -xtreg- has no option to supply them automatically.

Second, if you believe that yearly (or any time based) shocks (such 
as say national shifts in in the business cycle) that affect all of 
the relationships are present, you should include time-specific dummy 
variables.  But note that such models will be less efficient due to 
the increase in the number of parameters that must be estimated.  The 
larger t-statistic you mention probably stems from greater 
multicollinearity in a two- rather than a one-way fixed-effects model.

Dave Jacobs

At 03:29 PM 6/24/2008, you wrote:

> Dear Statalist,
>
> I have an old question, which command should I use for two-way fixed 
> effects? For example if I want to control both year and firm fixed 
> effects. I find two methods as follows:
> 1)
> . egen dummy = group(firm year)
> . xi: reg quantity price i.dummy
>
> and 2)
> . xi: reg quantity price i.firm i.year
>
> I find the two methods gave the same estimation of coefficient, but 
> method 2) seems to yield a large t-stat. Could someone please advise 
> which one should I use?
>
> Many thanks!
>
> Martin
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/